Arc-Flash Study

NOTE This standard governs the arc-flash study as an engineering deliverable: the engineer who performs it and when it is issued, the calculation method, the buses studied, the cases, the inputs the study adds to the system model it inherits, the clearing time it takes from the coordination settings, the content of the label at each bus, the PPE arc rating it hands to the facility's safety program, the mitigation it recommends, the report, and the review cycle that keeps the labels true. (1.1)

NOTE NFPA 70E 130.5 requires an arc flash risk assessment for every task a worker performs on energized equipment, and NFPA 70 110.16 requires the equipment likely to be worked on energized to carry a label that warns of the hazard; 110.16(B) adds, for service and feeder equipment rated 1200 A and more, the incident energy or the arc rating the worker needs. OSHA 29 CFR 1910.333 and 1910.335 require the employer to protect the worker from the hazard the equipment presents. None of these can be met from a nameplate; each rests on the incident energy at the bus, and this study is that calculation. (1.2)

NOTE The arc-flash study is the last of three studies performed on one system model. The short-circuit study under Short Circuit StudyShort-Circuit StudyResolves to the current adopted revision.sync/short-circuit-study gives the bolted fault current at every bus, the coordination study under Protective Coordination StudyProtective Device Coordination StudyResolves to the current adopted revision.sync/protective-coordination-study fixes the settings that decide how long the arc burns, and this study turns the two into incident energy, an arc-flash boundary, and a label. A change to either predecessor moves every result here, which is why this study is not started until the coordination settings are accepted and is re-run whenever they change. (1.3)

NOTE The following are governed elsewhere and are outside this standard: (1.4)

NOTE The study plan settles, before any case is run, the questions that otherwise surface when the labels are already printed: which buses are in and which are out, which operating states the worker will meet, whether the enclosure sizes are typical or measured, and whether the study will be back before the equipment is energized. A plan returned with the boundary drawn and the cases named is what makes the labels a record of the system rather than a best guess at it. (3.1.3)

NOTE The label on the door is read for the life of the equipment, and the five-year review and every re-study open the model that produced it. A review that has to rebuild the model from the PDF re-enters every enclosure size, gap, and working distance by hand, and a value that drifts in the re-entry prints on a new label as a fact. (3.2.3)

NOTE The engineer who performed the short-circuit and coordination studies already holds the model, the assumption register, and the settings the arc-flash calculation reads, and runs the arc-flash study as a third pass on the same file, which is why that arrangement is the one on nearly every project; the other four arrangements are the ones the short-circuit study offers for its own engineer, with the same trade-offs recorded under Short Circuit StudyShort-Circuit StudyResolves to the current adopted revision.sync/short-circuit-study, and a project selects one of them here where the arc-flash study is procured separately, as when an Owner engages an independent engineer to label an existing facility whose studies were performed by others. (4.1.2)

NOTE A preliminary issue on design data and preliminary settings is run while the equipment can still change, and it is the issue that decides whether a main section needs arc-resistant construction, a maintenance switch, or a faster upstream fuse before the switchgear is ordered; the final issue on the accepted coordination settings is the one every label is printed from; the confirmation against the as-left settings record compares the settings the study used with the settings the testing organization recorded at each device, re-runs the buses where they differ, and is the step that keeps a trip unit left at its shipping setting from invalidating a label that says otherwise. (4.2.2)

NOTE An arc-flash label is a statement about one device: that it will clear an arcing fault in the time the study read from its curve. A device set to a different value clears in a different time, and the label on the door is then wrong by a factor the worker cannot see. The sequence, settings accepted, settings applied, settings recorded, study confirmed, labels printed, is what makes the number on the door the number in the room. (4.2.7)

NOTE Commercial analysis software reads the arcing current off the device curve, applies the arcing current variation and the enclosure size correction of IEEE 1584-2018 at every bus without transcription, and keeps the arc-flash results in the same file the next re-study opens, which is why it is the tool on nearly every study; an independent calculation of the IEEE 1584-2018 equations, in a spreadsheet or by hand from the bolted fault currents and the clearing times tabulated in the two accepted reports, is reproducible by any engineer with those reports and is the tool a project accepts where the system is a single service with a handful of buses and the short-circuit study was itself a hand calculation, with the result that there is no model file to hand to the next review. (4.3.2)

NOTE The IEEE 1584-2018 model is empirical, built from a test program that spanned three-phase systems from 208 V through 15 kV, bolted fault currents from 500 A to 106 kA at 600 V and below and from 200 A to 65 kA above 600 V, electrode gaps from 6.35 mm to 76.2 mm at low voltage and from 19.05 mm to 254 mm at medium voltage, working distances of 305 mm and more, and five electrode configurations. Within that range it predicts the arcing current, the incident energy, and the arc-flash boundary from the bolted fault current, the voltage, the gap, the enclosure, the configuration, the working distance, and the arc duration; outside it, the equations are extrapolating from data they were never fitted to, which is why the study has to know, bus by bus, which side of the range it is on. (5.1.2)

NOTE The Ralph Lee method is a theoretical open-air calculation that overstates the incident energy at most buses, is the method the analysis software applies above 15 kV and on single-phase systems as a matter of course, and gives a conservative label at the cost of a conservative PPE requirement; extending the IEEE 1584-2018 model by judgment, as where a gap or a fault current lies just past the tested range, gives a realistic value at the cost of an assumption that the register has to carry and the reviewer has to accept; calculating nothing and labeling the bus to be de-energized before any work is the arrangement for a bus no one will ever open energized, as an outdoor transformer primary compartment, and it leaves the worker no number where the bus is opened anyway. The buses that commonly fall outside the range are the utility primary above 15 kV, single-phase equipment, and 208 V buses with a bolted fault current below 500 A. (5.2.2)

NOTE IEEE 1584-2002 permitted a study to leave out circuits below 240 V fed by a single transformer under 125 kVA, and a generation of studies and labels was built on that exemption. IEEE 1584-2018 withdrew it. The 2018 edition notes only that a sustained arc is less likely on a three-phase system at 240 V and below where the bolted fault current is below 2000 A, and that note is a statement about the likelihood of an arc, not an exemption from calculating its energy. The 208 V panelboard on a 75 kVA transformer is the bus a maintenance electrician opens most often, and it is calculated and labeled like every other. (5.3.2)

NOTE The IEEE 1584-2018 model is an AC model and offers nothing for a battery string or a DC bus. The maximum power method of NFPA 70E Informative Annex D.5 takes the arc at the point of maximum power transfer, half the bolted DC fault current at half the system voltage, needs only the bolted fault current and the clearing time, and overstates the energy at every bus, which is the trade a project accepts for a method anyone can check; the detailed arc model method of the same annex calculates the arc resistance from the gap and the current by the published arc models, gives a lower and more realistic energy, and needs a gap for every DC bus and an iteration the software performs. On a project with no DC equipment at or above 100 V, the threshold below which NFPA 70E assigns no DC arc-flash PPE, the question does not arise, and the study states so. (5.4.2)

NOTE NFPA 70 110.16(A) names the equipment that carries an arc-flash warning, switchboards, switchgear, panelboards, industrial control panels, meter socket enclosures, and motor control centers, and then adds any other equipment likely to be worked on energized, and the list above is that clause written out for the equipment a project actually installs. A transformer terminal compartment and a current transformer cabinet sit on the line side of a main and are among the highest-energy buses in a system; a motor disconnect and a busway plug-in unit are opened energized more often than any switchboard; a junction box with a splice is the one class the project adds where its maintenance practice opens boxes under load, and leaves out where it does not. The classes a project leaves out are the ones no worker opens energized. (6.1.2)

NOTE The main breaker protects everything below it and nothing above it. An arcing fault on its line terminals, in the incoming compartment, or on the bus of a transformer secondary, is cleared by the device on the primary side of the transformer, which sees the arcing current reduced by the turns ratio and often sits on its long-time curve for seconds. That is why the main section of a switchboard is routinely the highest-energy bus in a building, and why a single label for the section states the line-side value. (6.2.3)

NOTE A lower fault current does not mean a lower incident energy. The arcing current falls with the bolted current, and when it falls below the instantaneous pickup of the upstream device the clearing time jumps from a few cycles to the time-delay band or the long-time curve, and the energy rises with it. The maximum case with the arcing current variation catches that effect over the variation the model expects at one bolted current; the minimum case, with the utility at its minimum contribution and the motors and the smallest source configuration, catches it where the utility's own range is wider than the variation, as on a service fed from a weak or switched utility system. A project runs the maximum case alone where the short-circuit study delivered no minimum. (7.1.2)

NOTE The arcing current variation replaces the fixed 0.85 multiplier of the 2002 edition with a correction that depends on the voltage, the gap, the configuration, and the fault current, and it is the step most often found missing in a study converted from a 2002 model by changing the edition setting. A bus whose reduced arcing current sits below a pickup that the full arcing current clears is a bus with two labels' worth of difference between the two runs. (7.2.3)

NOTE A generator alone delivers less fault current than the utility and clears more slowly, because its decrement drops the arcing current into the time-delay bands of the devices below it, and the incident energy on generator power is routinely higher than on utility power at the same bus; the utility and the generation paralleled gives the sum of both at every bus between them and is the case for the paralleling bus and the main switchgear during a closed transition or in extended parallel operation; a tie closed between two services or two transformers doubles the source behind the common bus, and is a case whether or not the operating procedure forbids it, because the tie can be closed; a UPS or an inverter in bypass passes the upstream source through to loads that on inverter power see only the current limit; the maintenance bypass of paralleling switchgear or of a transfer switch puts a bus on a source and a clearing device it is never on in normal operation, and is the case the paralleling switchgear standard requires on the bypassed distribution's label. A case left out is a configuration in which a worker can open a door against a number no one has calculated. (7.3.2)

NOTE Both values on the label let the worker see what the switch buys and choose the PPE for the state the switch is in, at the cost of a label with two numbers that a worker in a hurry can read wrong; the maintenance-state value alone with the instruction to engage the switch first is the arrangement where the Owner's program makes engaging the switch a step of every energized task at that equipment and the switch's status light is visible with the door closed, and it leaves the worker unprotected on the day the switch is bypassed or failed; the normal-state value alone on the label treats the switch as a bonus the label does not count on, gives the most conservative PPE, and loses the reduction the switch was bought for. (7.4.2)

NOTE The maintenance switch, the interlock, the active mitigation system, and the arc-resistant enclosure are furnished and tested under the equipment standards; this study calculates what each does to the energy at the bus and tells the worker which of them the label counts on. A label that quietly assumes an interlock that was never commissioned is the failure the per-bus statement of dependence exists to prevent. (7.4.6)

NOTE An arc-flash study collects four inputs the system model does not already carry, the electrode configuration, the enclosure dimensions, the bus gap, and the working distance at each bus, and everything else it reads is already in the model it inherits. A study that re-collects the utility letter and re-enters the transformer impedances is building a second model, and the two will differ by the time the labels are printed. (8.1.3)

NOTE IEEE 1584-2018 tests five configurations, vertical conductors in a box (VCB), vertical conductors terminating in an insulating barrier in a box (VCBB), horizontal conductors in a box (HCB), and vertical and horizontal conductors in open air (VOA and HOA), and at the same bus the configuration alone moves the incident energy by a factor of two or more, with VCBB and HCB above VCB in a low-voltage enclosure. Setting the configuration per bus, from the compartment the worker actually opens, gives the value the equipment presents, at the cost of a survey of each equipment class and a judgment where a compartment holds more than one geometry; VCB for every enclosure and VOA in open air is the setting the software ships with and the one most studies run, and it understates the energy at a bus whose exposed conductors are horizontal or terminate on a barrier, as the line terminals of a panelboard main and the bus stabs of a motor control center; the most severe enclosed configuration for each voltage class overstates the energy at every bus with a vertical geometry, which is most of them, and buys a conservative label at the cost of PPE and of mitigation recommended where none is needed. (8.2.2)

NOTE The typical enclosure dimensions IEEE 1584-2018 tabulates for each equipment class are the values nearly every study applies, because the correction factor is modest over the range of real enclosures and a measured dimension changes the energy at a bus by less than the uncertainty in its clearing time; the dimensions of the equipment furnished, taken from the shop drawings or measured at an existing item, are the basis a project selects where a compartment is far smaller than its class, as a shallow motor disconnect or a small junction box, where the correction factor rises with the smaller volume and the typical value understates the energy. (8.3.2)

NOTE The typical gaps IEEE 1584-2018 tabulates by equipment class and voltage, 25 mm for a low-voltage panelboard and motor control center, 32 mm for low-voltage switchgear, 104 mm for 5 kV switchgear and 152 mm for 15 kV switchgear, are the values nearly every study applies, because the gap enters the arcing current and the energy weakly over the range any one class spans; the gap of the equipment furnished, from the manufacturer's data or a measurement, is the basis a project selects at medium voltage, where the range across designs is wide and a 15 kV compartment built to a 104 mm gap is calculated at a 152 mm gap otherwise. (8.4.2)

NOTE The working distance is the distance from the arc to the worker's face and chest, and the incident energy falls with roughly the square of it, so the distance decides the label as much as the clearing time does. The three defaults, 18 in for a panelboard or a motor control center, 24 in for low-voltage switchgear and switchboards, and 36 in for medium-voltage switchgear, are the typical distances IEEE 1584-2018 tabulates and the ones the label is read at; a project shortens the distance at an equipment class where its tasks put the worker's face closer, as a panelboard mounted so that the worker crouches at it, and lengthens it only where every energized task at the class is performed with a tool that keeps the worker further away. (8.5.4)

NOTE The curve says when the trip unit decides; the arc ends when the contacts part and the arc inside the breaker goes out. For a low-voltage power breaker the difference is about three cycles, for a molded-case breaker about one, for a medium-voltage breaker three to five, and at a bus cleared on an instantaneous element it is most of the duration. A study that reads the curve and stops has every instantaneous bus wrong by a factor. (9.1.8)

NOTE IEEE 1584-2018 and NFPA 70E Informative Annex D note that a worker exposed to an arc is likely to have moved out of its reach within about two seconds, and nearly every study caps the duration there at a bus cleared on a long-time curve, because the uncapped value, often hundreds of calories, is an energy no PPE addresses and the cap is what makes the label a number a worker can act on; calculating the full clearing time leaves the label with that energy, which is the honest statement at a bus from which the worker cannot step away. (9.2.2)

NOTE NFPA 70E 130.5(F) permits either the incident energy analysis method or the PPE category method on a given item of equipment and forbids both on the same one. The incident energy method prints the calculated energy at the working distance and is the method on every project that has performed this study, because the study exists to produce the number; the PPE category method prints a category from the tables of NFPA 70E 130.7(C)(15), which apply only to the equipment types, fault currents, and clearing times the tables list, and is selected where the Owner's program is built on categories and every bus falls within the table parameters. A label that prints a category beside a calculated energy invites the worker to pick whichever reads lower. (10.1.2)

NOTE The voltage, the arc-flash boundary, and the energy or the category are the three things NFPA 70E requires on the door, and the rest is what makes the label usable without the report: the working distance tells the worker what the energy was calculated at, the shock boundaries from NFPA 70E Table 130.4 put the approach limits where the arc-flash boundary is, the designation ties the door to the one-line and the schedule, the clearing device tells the next engineer what the value depends on, and the date tells the reviewer whether the label is current. A study report reference is added where the Owner's program requires the worker to be able to find the calculation; a site-specific PPE level is added where the Owner's program names its levels. (10.2.4)

NOTE ANSI Z535.4 reserves DANGER for a hazard that will cause death or serious injury if not avoided and WARNING for one that could, and NFPA 70E leaves the threshold between them to the facility. Forty calories per square centimeter was for many editions the line above which NFPA 70E assigned no PPE category, and it remains the threshold most programs use for DANGER and for the rule that the bus is not worked energized, which is why it is the basis on nearly every study; setting the threshold at the arc rating of the highest PPE the Owner stocks puts DANGER exactly where the Owner's workers have no protection, which on a program stocked to 40 cal/cm² is the same line and on one stocked to 65 or 100 cal/cm² is a higher one; WARNING on every label reads the signal word as a hazard class rather than a severity and leaves the worker to read the energy, which is the practice of a program that treats every energized task alike. (10.3.2)

NOTE The study produces the energy and the arc rating that meets it; the Owner's program decides which garments, which face protection, and which tasks go with each rating, and that division is why the table states the rating as a number and not as a garment. A site level is a convenience for a program that stocks three or four kits, and it is only as good as the arc rating the Owner attaches to it, which is why the Owner furnishes the ratings before the study assigns the levels. (10.4.4)

NOTE A threshold of 40 cal/cm² confines the recommendations to the buses the Owner's program will not work energized and is the arrangement where the aim is to make every bus workable; 25 cal/cm² adds the buses that require the heaviest PPE a program commonly stocks; 8 or 12 cal/cm² adds the buses above the daily-wear arc rating of most programs and is the threshold of an Owner whose aim is to perform routine tasks in daily wear; 1.2 or 4 cal/cm² asks for a recommendation at nearly every bus and is the arrangement of a study procured specifically to design a mitigation program. The threshold moves the length of the recommendation list and not the results. (11.1.2)

NOTE Every option on the list shortens the arc, moves the worker, or contains the blast, and each has a place. A revised setting costs nothing and is bounded by the selectivity the coordination study established; a maintenance switch cuts the energy only while a worker remembers to turn it on; zone-selective interlocking cuts it at every bus below the interlocked device at all times and needs wiring between the devices; differential protection clears a bus fault in a few cycles regardless of the arcing current and costs current transformers and a relay; a current-limiting fuse ahead of the bus ends the arc in a quarter cycle where the arcing current is in its current-limiting range and does nothing where it is not; an optical relay trips on the light of the arc in milliseconds; arc-resistant construction contains the arc with the doors closed and changes nothing with a door open; remote racking and operation move the worker outside the boundary for the tasks that start most arcs; relocating a bus below a faster device or a transformer is the design change that fixes the problem at the one-line. The options a project leaves off the list are the ones its equipment and its budget rule out before the study begins. (11.2.2)

NOTE The hierarchy of controls in NFPA 70E puts elimination of the energized task first and PPE last, and the mitigation list is the middle of that hierarchy: the engineering controls that lower the energy a worker meets when the task cannot be eliminated. A study that reports the energy and stops has told the Owner where the problem is; one that prices the options has told the Owner what to do about it, and that is what the recommendation list is for. (11.2.6)

NOTE The schedule is the handoff. The report is read by the engineer who reviews it; the schedule is read by the printer who makes the labels and by the electrician who installs them, and it has to carry everything they need in one row per door, so that no one is reading a value off a results table at the print shop. (12.3.4)

NOTE The sealed PDF is the record, the native model is what the five-year review opens, the editable schedule is what the labels are printed from, the editable one-line is what the Owner's drafting staff fold into the record drawings, and the printed copy is what sits in the electrical room with the safety program; a project adds the last two where its record-drawing and closeout practice call for them. (12.4.2)

NOTE Five years is the longest interval NFPA 70E permits between reviews and the one nearly every program adopts, on a facility whose change procedures bring every trigger event to the responsible engineer as it happens; a shorter interval is the arrangement of a facility that is reconfigured often, that has no change procedure capturing electrical work, or whose utility reinforces its system on a schedule the Owner does not see, and it buys a label that is checked before the system has drifted from it. (13.2.2)

NOTE The arc-flash label inherits every trigger of the two studies beneath it and adds its own, and most of them are ordinary maintenance: a trip unit replaced with the one in stock and left at its shipping setting, a maintenance switch bypassed to stop a nuisance trip, a panel replaced with a deeper one. None of them look like a reason to commission a study, which is why the list is written for the operations staff in the words of the work they do, and why the label carries its date. (13.2.7)